Compare grantee reports over time without confusing changes in outcomes, coverage or definitions. Build a source-linked history with a worked cohort example.
Analyze grantee reporting over time by linking each report to the same grantee and relevant grant, while preserving its period, cohort, definitions and source version. Compare like measures, show missing information and changes in coverage, and read the narrative alongside the numbers. A longer history helps explain change only when the underlying records remain comparable; several years of files do not automatically form a reliable trend.
By Sopact Academy · Reviewed September 12, 2026. This lesson combines cited public guidance with practical workflow recommendations. Figures and teaching scenarios are illustrative unless explicitly identified as a published case.
Grant Intelligence · Chapter 17 · Analyze
Bring: two or more reports, the relevant definitions and the financial context from chapter 16.
Leave with: a period-by-period evidence table, a list of comparability limits and one supported explanation of what changed. The numerical example below is fictional.
A grantee history, a grant history and a participant history are different. One organization can run several grants and serve a new cohort each year. A stable organization ID does not make those participants the same people.
State the level in the question. “Did the grantee report better coverage this year?” can be answered from annual summaries. “Which participants stayed employed?” requires suitable linked person-level observations and definitions of retention.
For every observation, retain the grantee ID, grant ID, report period, measure definition, cohort and source. Keep the original submission date separate from the period the report describes.
If an organization changes its name, verify the relationship rather than creating a duplicate history or merging on name alone. A merger, a new program or a changed grant scope may require a documented break in the series.
In a configured Sopact workflow, source-linked records can keep new reports associated with earlier context. The value comes from those relationships and definitions, not just putting all the documents in one folder. Confirm what history is available and authorized before asking a question across it.
The Office for National Statistics’ quality guidance treats comparability across time and domains as a dimension of data quality, alongside accuracy, relevance and other considerations. Apply that principle to the reports: a consistent-looking chart is not enough if the measures changed.
| Check | Question | What to record |
|---|---|---|
| Measure | Does “employed” mean the same thing in each report? | Definition and version |
| Timing | Is follow-up at the same point after completion? | Observation window |
| Population | Is this the same cohort or a new one? | Cohort and inclusion rules |
| Coverage | How many people or records have known results? | Numerator, denominator and missing share |
| Method | Did survey wording, collection channel or evidence source change? | Method change and likely interpretation limit |
| Financial scope | Are period, basis, currency and grant share comparable? | Finance-approved scope and any restatement |
When a definition changes, preserve both versions. If a valid common measure can be reconstructed, document the method and keep the original values. If not, mark the break and avoid presenting the series as directly comparable.
Consider two different annual cohorts from the same fictional grantee. Both have 150 starters. In the first cohort, employment status is known for 100 people and 60 are employed. In the second, status is known for 120 and 72 are employed.
| Measure | Cohort A | Cohort B |
|---|---|---|
| Starters | 150 | 150 |
| Known employment status | 100 | 120 |
| Reported employed | 60 | 72 |
| Employed among known results | 60÷100 = 60% | 72÷120 = 60% |
| Follow-up coverage | 100÷150 = 66.7% | 120÷150 = 80% |
| Unknown status | 50 | 30 |
The number of confirmed employment outcomes increased, while the rate among people with known status remained 60%. Coverage improved by 13.3 percentage points. That is useful progress in reporting coverage, but it does not by itself establish that participants were more likely to become employed.
These are different cohorts. The comparison does not show individual change or employment retention. Even if definitions and follow-up timing match, differences in the people served or external conditions may matter.
Review the narrative for changes in delivery, staffing, participant needs, access barriers and data collection. Tie each explanation to its period and source. A later report may clarify an earlier problem, but preserve what was known at the time.
Distinguish the grantee’s explanation from your interpretation. If the grantee says follow-up coverage improved after adding phone calls, record that account. Do not state that phone calls caused a change in employment unless the evidence supports that claim.
Changes in report length or writing style can also affect what is mentioned. A longer narrative may contain more barrier descriptions without indicating that barriers became more common. Keep coding definitions and the available text context visible.
Use an instruction that names the period range, grant scope and relevant measures. Require the answer to identify gaps and definition changes before summarizing a trajectory.
Compare the supplied reports for [grantee and grant] across [periods]. Use the attached measure definitions and identify any changes in cohort, timing, coverage, method or financial basis. For each comparison, cite the source period and record. Separate changes in reported values from changes in data availability. Keep the grantee’s explanation distinct from your interpretation. If a series is not comparable, state why rather than drawing a trend line. Do not infer individual change from different annual cohorts.
Check the answer against the underlying records. If an older file is missing or unreadable, the assistant should identify the gap rather than supply a smooth narrative across it.
Watch the explainer · 6 minutes 7 seconds
This Sopact video explains why sources, definitions and audience context matter even when the numbers are clean. Apply that lesson when checking whether changes in reported results reflect outcomes, coverage or a different definition.
Watch on YouTube if playback is unavailable · Browse the video library
Alongside the numerical table, record changes that affect interpretation: an approved target revision, a new measure definition, a service redesign or a change in collection method. Keep the event date and the reporting period distinct.
Use plain labels such as “definition changed,” “follow-up not yet due” or “source revised.” Do not imply an outcome improved merely because the data became more complete.
If an earlier report is corrected, retain the original and corrected version, with the reason. A reader should be able to reproduce both the earlier published view and the current corrected view where the reporting process requires it.
Two points show a difference between two observations. They may not establish a persistent trend. A plateau or decline needs a defined measure, adequate observations and an interpretation of context.
Do not fill missing years with invented values to create continuity. If you use an estimate under an appropriate analytical method, label it and retain the basis; it must not look like a reported observation.
For portfolio summaries, do not average grantee percentages without examining their denominators and whether the measures are comparable. Explain which records are included and which are excluded, rather than hiding the exclusions in a total.
Take three fictional reports. Keep the first two measure definitions consistent. In the third, change the follow-up window or definition of the outcome. Ask a colleague to identify which comparisons remain valid.
Then add a corrected source file for the first period. Confirm that the change is visible, earlier values are not silently overwritten and the summary explains what was revised.
Pass condition: the reader can distinguish a change in results, a change in coverage and a change in measurement. Every important statement can be traced to the relevant period and source.
It means examining a grantee’s reports across time while preserving the grant, period, definitions and sources. State whether the analysis follows an organization, grant, cohort or the same individuals.
No. You must also check the measures, populations, timing, coverage and methods. A stable organization ID does not mean the same participants were measured.
Keep both versions and document the change. Reconstruct a common measure only where the evidence supports it; otherwise mark a break rather than implying direct comparability.
Yes. More complete follow-up can increase the count of known outcomes without changing the rate among known results. Show coverage and missing data alongside the outcome.
It can assist in a configured workflow with the relevant authorized sources. Verify its comparisons and require it to identify missing periods, definition changes and unsupported explanations.
No. They show a difference between two observations. A persistent trajectory needs suitable measures, observations and interpretation of context.
Do not invent reported values. Keep gaps visible. Any appropriate estimate must be clearly labeled with its method and limitations.
Carry the comparable series and its limits into renewal, learning or portfolio discussions. Next, examine what a U.S. Form 990 can and cannot establish in a grant review. Keep this timeline available when combining that filing with other evidence.
For presenting the result, use the impact report writing ebook and example reports. Return to the Grant Intelligence course to continue.
Bring two reporting periods and their definitions. Explore how to keep the history and source evidence connected for a reliable comparison.
Explore Applications & Grants →
